You are on page 1of 8

Electrochimica Acta 54 (2009) 46964703

Contents lists available at ScienceDirect

Electrochimica Acta
journal homepage: www.elsevier.com/locate/electacta

Crevice corrosion cathodic reactions and crevice scaling laws


Glyn F. Kennell, Richard W. Evitts
Department of Chemical Engineering, University of Saskatchewan, 57 Campus Drive, Saskatoon, SK, Canada S7N 5A9

a r t i c l e

i n f o

Article history:
Received 12 December 2008
Received in revised form 27 March 2009
Accepted 27 March 2009
Available online 7 April 2009
Keywords:
Stainless steel
Polarization
Modeling studies
Crevice corrosion
Scaling laws

a b s t r a c t
A numerical model that predicts the rates of metal dissolution and electrolyte composition along the
length of a metallic crevice was used to simulate the crevice corrosion of AISI 304 stainless steel. The
model considers both the forward and reverse electrochemical reactions that might take place during
the corrosion process. The environment exterior to the crevice, where a net cathodic current is produced,
was simultaneously modeled. It was found that cathodic reactions are likely to occur towards the tip of
the crevice. For the case when the hydrogen evolution reaction was considered as a possible cathodic
reaction in the crevice, it is shown how the delayed instigation of this reaction may be the cause of an
experimentally observed increase in pH at the crevice tip. Two critical crevice scaling laws were examined using model predictions and one scaling law t the model predictions very well. This scaling law
differentiates between crevices that will undergo active corrosion and those that will remain indenitely
passive.
2009 Elsevier Ltd. All rights reserved.

1. Introduction
Crevice corrosion is a localized form of corrosion that can cause
considerable damage. It can occur on passive metals under conditions that do not normally cause high levels of uniform corrosion.
These facts, combined with the fact that crevice corrosion can
be difcult to detect, may lead to situations where the structural
integrity of equipment is challenged by crevice corrosion under
conditions that corrosion was not expected. The damage presented
by crevice corrosion creates the need for a detailed understanding of the complex phenomena causing this type of corrosion. This
understanding would help enable the prediction and prevention of
crevice corrosion, as well as provide insight into related forms of
localized corrosion, such as stress corrosion cracking.
Crevice corrosion occurs in occluded spaces, possibly caused by
a ange or under sediment deposited on the inside of a pipe. In a
crevice a microenvironment may form in which the concentrations
of species may be considerably different from those in the bulk
electrolyte. A concentration gradient between the crevice and bulk
electrolyte can be caused by a very low rate of corrosion that manifests as a small leakage or passive current, which is always present
with passive metals. A passive current density is often much less
than 1 A/cm2 .
For metal surfaces exposed to the bulk electrolyte, the build up of
species due to the passive current is overwhelmed by mass transfer
of the species away from the surface. Therefore, the concentrations

Corresponding author. Tel.: +1 306 966 4766; fax: +1 306 966 4777.
E-mail address: Richard.Evitts@usask.ca (R.W. Evitts).
0013-4686/$ see front matter 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.electacta.2009.03.080

of species in proximity to the metal are similar to bulk electrolyte


concentrations. However, for a stagnant electrolyte within a crevice
mass transfer is limited by the width and gap of the crevice mouth
through which all species must traverse. This mass transfer limitation allows the build up and depletion of species in the crevice
electrolyte due to the passive current. The oxygen in the crevice may
become depleted if oxygen reduction occurs faster than the transport rate of oxygen through the crevice mouth. Depletion of oxygen
in a crevice is the rst stage of crevice corrosion and this may be
followed by an incubation period during which the crevice acidies
due to metal ion hydrolysis. At the end of the incubation period the
passive lm on the crevice will be damaged. Conversely an innite
incubation corresponds to an absence of crevice corrosion.
After the oxygen in the crevice becomes depleted the reduction of oxygen continues on the bold surface and anodic reactions
may still occur in the crevice if they are supported by bold surface
cathodic reactions. These cathodic reactions can support crevice
anodic reactions if ions and electrons can move between the
cathodic and anodic areas. A complete electrical circuit is developed with aqueous ionic charge transfer and electronic conduction
in the solid phase. The conduction of charged ions through the electrolyte may cause a signicant drop in potential, commonly referred
to as the iR drop [1].
Chloride anions also migrate into the crevice, attracted to the
anodic metal surface. The increased concentrations of hydrogen and
chloride ions in the crevice attack the passive lm protecting the
metal and cause the passive current to increase. An increase of corrosion current through the crevice increases the potential drop into
the crevice if the change in conductivity of the electrolyte is negligible. With this decrease in potential some regions in the crevice

G.F. Kennell, R.W. Evitts / Electrochimica Acta 54 (2009) 46964703

may be at an electric potential that causes signicantly higher rates


of corrosion, viz for a passive metal this potential may be below the
critical potential at some regions in the crevice. These regions would
be in a state of active crevice corrosion and signicant corrosion
within the crevice may be occurring.
An actively corroding crevice undergoes corrosion at rates that
can be several magnitudes larger than when the metal is in the
passive state. These high rates of corrosion can further increase concentrations gradients within the crevice and change the dimensions
of the crevice. These changes may cause the rate of active corrosion
to change and relocate or expand the active region. These changes
occur during the propagation phase of crevice corrosion.
During the propagation phase, rates of corrosion near the crevice
mouth may become very large. These large rates of corrosion cause
large electric potential drops in the electrolyte close to the crevice
mouth mainly due to larger current ow in the solution. If the
potential drop near the crevice mouth becomes too large, cations
produced at active anodic areas deeper in the crevice will be less
likely to traverse this large potential barrier and exit the crevice.
Under these circumstances electroactive regions deeper into the
crevice may cease to produce a net anodic current. This would result
in no net metal dissolution at these deeper regions, whilst regions
close to the crevice mouth may undergo high levels of corrosion.
These phenomena are explored in the model presented in this paper
through the explicit representation of forward and reverse reactions for the corrosion of AISI 304 stainless steel and the hydrogen
evolution reaction.
Crevice corrosion can be classied due to its ability to initiate
with or without an induction period. Crevice corrosion can therefore be referred to as immediate or delayed crevice corrosion [2].
Immediate crevice corrosion occurs when the potential at some
area in the crevice is immediately forced below the crevice corrosion critical potential upon oxygen depletion. For this to occur,
the passive current emitted along the crevice must generate a sufciently large iR drop in the solution at bulk pH and chloride ion
concentrations to cause the potential of the metal to fall in the active
region. For the situations where an incubation period occurs before
active corrosion, higher levels of hydrogen and chloride ions will
build up in the crevice. These species will attack the passive lm and
cause a higher passive current density, which in turn will increase
the iR drop in the crevice.
1.1. Mathematical models of crevice corrosion
A number of models describing crevice corrosion have previously been presented in the literature. The rst model was
developed by Oldeld and Sutton [3]. It was a general model of
crevice corrosion that predicted the incubation period of crevice
corrosion. This model considered diffusion and migration and used
empiricisms that simplied aspects of the simulation, including the
mass transport of species into the crevice. The model assumed the
determining factor for initiation of active crevice corrosion was the
aggressive levels of pH and chloride ion concentration in the crevice
attacking the passive lm. This theory is commonly called Critical Crevice Solution Theory (CCST). The iR drop mechanism was
rst published by Pickering [1], who claimed that the initiation of
crevice corrosion was dependent upon the potential drop between
anodic active regions in the crevice and cathodic regions on the bold
surface. Xu and Pickering [4] developed a model that focused upon
the difference in potential between the bold and interior crevice
surfaces. This model predicted electrical effects inside the crevice
without modeling any mass transfer or concentration effects other
than the conductivity of the electrolyte. This model was used to
simulate an iron crevice immersed in an ammoniacal electrolyte.
Results from simulations indicated that the potential applied to
the bold surface inuences the potential and current distributions

4697

within the crevice. Watson and Postlethwaite [5] developed a new


CCST model for mass transport processes occurring in a crevice and
applied it to simulating the incubation period of crevice corrosion
of stainless steels. This represented a more rigorous representation of the transport processes occurring in the crevice based on
innitely dilute solution theory. Walton et al. [6] developed a transient general model for crevice corrosion. This model included the
inuence of iR effects on the transport of species in the electrolyte;
however, electric potential effects on the rates of electrochemical
reactions were not considered when they simulated the experiment
conducted by Alavi and Cottis [7], as a passive current was assumed.
The predicted pH was close to the average pH seen experimentally, but the experimental pH prole was not predicted. Engelhardt
et al. [8] developed a model for crevice corrosion in a Pressurized
Water Reactor Steam Generator that considered mass transport due
to diffusion, migration, and convection. This model considered iR
drops inside and exterior to the crevice; however, electrochemical reactions for the dissolution of the metal were considered
independent of potential inside the crevice, with the corrosion current empirically related to the pH of the electrolyte. The interior
of the crevice was coupled with the exterior of the crevice using
the principle of charge conservation. White et al. [9] used CCST to
determine the onset of active crevice corrosion for the Alavi and
Cottis [7] experiment. Potential differences were calculated using
the NernstPlanck equation and were assumed xed throughout
the simulation. The crevice was numerically coupled to a spherical
external environment, which was assumed to have zero gradients
of concentration and potential. Passive current independent of electric potential was assumed within the crevice. The model did not
predict a pH in the crevice that would cause active crevice corrosion.
However, an empirical correlation was determined that described
the corrosion current within the crevice and this provided results
that better t the experimental results of Alavi and Cottis [7]. Since
the empirical correlation was based on Tafel behaviour, White et al.
[9] concluded that the corrosion current within a corroding crevice
may be described using Tafel-like dissolution kinetics. Cui et al. [10]
developed a mathematical model describing the cathodic area supplying a net cathodic current to a corroding crevice with an electric
potential at the crevice mouth equal to the repassivation potential.
Potential gradients were calculated using Laplaces equation. The
cathodic current was assessed using Tafel kinetics with mass transfer limitations. Heppner [11] developed a crevice corrosion model
that rigorously described the transport of species within a corroding
crevice by utilizing Pitzers ionic interaction model. Chemical reactions occurring in the crevice were also rigorously modeled through
the implementation of a large number of chemical equilibria. Simulations were conducted for the experiment of Alavi and Cottis [7],
and the average experimental pH was accurately predicted, but the
pH prole was not. Kennell et al. [12] presented a model for crevice
corrosion that combined CCST and iR drop theories. It considered
the mass transport of species and the electric potential drops both
interior and exterior to the crevice. The two environments were
coupled using the principle of charge conservation. Mass transport
inside the crevice was based on innitely dilute solution theory and
considered transport due to diffusion, migration, and electroneutrality. Electrochemical reactions for the dissolution of the alloy and
reduction of oxygen were considered dependent upon the electric
eld close to the metal surface. The reactions were modeled using
Tafel kinetics. The experiment of Alavi and Cottis [7] was simulated and the experimental results, including the pH prole along
the length of the crevice, were predicted very well. However, Tafel
kinetics for metal dissolution poorly represented the phenomena
occurring towards the crevice tip, where the potential approached
the reversible potential of the alloy and the reverse reaction became
signicant. Other cathodic reactions that were neglected include
the hydrogen evolution reaction.

4698

G.F. Kennell, R.W. Evitts / Electrochimica Acta 54 (2009) 46964703

2. Theory
A mechanistic model is presented that describes a metallic
crevice in various stages of corrosion. At the beginning of the simulation it is assumed that there are no concentration gradients in the
electrolyte in the crevice or in the bulk solution. The passive current that exists at the prescribed initial conditions is assumed to
exist along all metal surfaces and causes the depletion of oxygen in
the crevice. During the incubation period, changes in concentration
gradients, passive currents, and electric potentials are predicted.
If active corrosion is predicted the model determines the rate of
active corrosion within the crevice. During the propagation phase
the iR drop may cause the potential drop between anodic crevice
areas and cathodic bold areas to increase [1]. If this potential drop
becomes too great some anodic regions in the crevice will decouple
from cathodic areas on the bold surface. It is assumed (and veried
in Section 3 of this paper), that decoupled anodic regions do not
produce a net anodic current that is transported to the bold surface. This model presented here is unique as it is capable of (a)
describing the phenomena occurring during the decoupling process, and (b) possible cathodic (and anodic) reactions occurring at
decoupled regions. This includes the possibility for net cathodic
reactions occurring at decoupled regions towards the crevice tip.
2.1. Mathematical model development
Some aspects of the overall model that have been previously
published will be briey presented here for completeness [5,1218].
Kennell et al. [12] describe in detail the development of an earlier
version of the model presented in this paper. The model presented here calculates the corrosion current in an AISI 304 stainless
steel crevice by considering both the forward and reverse electrochemical reactions, instead of assuming simple Tafel kinetics.
This greatly simplies the numerical solution of the mathematical
model because a continuous function is now used to describe the
rate of substrate dissolution rather than the Tafel equation, which
was not assumed valid at potentials less than the reversible potential in the previous model. This also increases the exibility of the
model as the assumption of zero current at decoupled regions is
no longer applied [12]. This allows for an inspection analysis of
the decoupling process of anodic sites in the crevice from cathodic
sites on the bold surface. The model also includes a method for
representing the hydrogen evolution reaction (HER).
Mass transport within the crevice is determined using Eqs. (1)
and (2). Eq. (1) describes the transport of species due to convection,
diffusion, diffusion potential, and electromigration [5]:
Ci
zi ui Ci
zi ui Ci 
= 
i 
zj Dj Cj + Di 2 Ci + S
t
F
zj2 uj Cj
zj2 uj Cj j=1
k

(1)

Eq. (1) was solved over a one-dimensional grid using a hybrid


CrankNicolson method [17]. The boundary conditions associated
with this equation were zero ux at the crevice tip and bulk solution
concentration outside the crevice mouth. The values for diffusion
coefcients and mobilities are found elsewhere [16]. Near electroneutrality of the electrolyte in the crevice is modeled by applying
the algebraic correction term developed by Heppner and Evitts [18]
to Eq. (1):
F
Ci = Ciold zi ui Ci


(2)

The source term, S, in Eq. (1) represents species produced or


depleted by chemical reactions. For chemical reactions kinetic

Table 1
Values for parameters used in simulating an AISI 304 stainless steel corroding
crevice.
Anode

i0,a
1 102pH A/cm2

E0,a
0.259 V/SHE

TS
0.03 V/decade

Ecrit
0.009 V/SHE

Cathode

i0,c [10]
1 109 A/cm2

E0,c [10]
0.191 V/SHE

TS [10]
0.1 V/decade

Ecorr
0.1 V/SHE

behaviour is neglected [15]:

Cjvij Ki = 0

(3)

where Cj is the equilibrated concentration of the jth species occurring in the ith chemical reaction, and K is the equilibrium constant.
A mass balance equation can be written for each species involved
in chemical reactions [15]:
Cj = Cjinit +

n


Xi vji

(4)

i=1

The mass balance equations are combined to provide a guaranteed non-singular matrix. This matrix was solved using the
NewtonRaphson method.
Mass transport exterior to the crevice was modeled using
the equation developed by Kennell et al. [12] for an oxygenated,
convection-free electrolyte:
C
2 C
i
KMT
=D 2 +
([Bulk] [Surface])
y
zFz
t
x

(5)

Eq. (5) considers the diffusion of oxygen parallel to the bold surface, the transfer of oxygen between the bulk electrolyte and bold
surface and the consumption of oxygen at the bold surface due
to electrochemical reactions. This equation was solved implicitly
over a one-dimensional grid parallel to the bold metal surface
using central space discretization. The initial condition was bulk
concentration at all nodes and boundary conditions were zero concentration gradient at the crevice mouth and bulk concentration
at the node furthest from the crevice. The diffusion coefcient of
oxygen was assumed to be 5.5 108 m2 /s and the electrolyte was
assumed to be under atmospheric conditions.
For oxygen depleted areas in the crevice with surface overpotential greater than the critical potential the passive current was
determined by modifying a reference current for pH using a Freundlich adsorption equation [19]:
log (iP ) = log (k) npH

(6)

For all crevice areas with a surface overpotential less than the critical potential the rate of corrosion was determined by considering
both the forward and reverse reactions:
i = i0,a (10(TS (a E0,a )) 10(TS (E0,a a )) )

(7)

The critical potential, the Tafel slope, the reversible potentials and
exchange current density values used in this model are given in
Table 1. The electric potential gradient was calculated in the crevice
using [20]:

i
F
zj Dj Cj




(8)

and the surface overpotentials were determined with reference to


the potential at the crevice mouth [12]:
a = Emouth +

(9)

The electric potential gradient parallel to the bold surface was calculated using [12]:

i
RT

ln

2F

 [O ]

2 mouth
O2

(10)

G.F. Kennell, R.W. Evitts / Electrochimica Acta 54 (2009) 46964703

4699

where the reduction of oxygen was assumed to adhere to Tafel


behaviour and the oxidation of metal was assumed to occur in the
passive region (ip ) and calculated via Eq. (6) [12]:
i = i0,c 10(TS (E0,c c )) ip

(11)

The values for the exchange current density and reversible potential
can be found in Table 1. The surface overpotential was again determined by using the potential at the crevice mouth as a reference
[12]:
c = Emouth

(12)

The hydrogen evolution reaction was considered for some of the


simulations conducted and the contribution to the current density
determined by
iH = i0,H 10(10(E0,H a ))

(13)

where the exchange current density for the hydrogen reaction was
assumed to be equal to that on iron and given by [21]:
i0,H = 1 105.9 (10pH )

0.5

(14)

and the reversible potential was calculated using the Nernst equation:
E0,H = 0.0591pH

(15)

2.2. Numerical solution


The numerical solution of the mathematical model is similar to
the method employed by Kennell et al. [12]; however, the implementation of Eq. (7) in the current model instead of simple Tafel
behaviour simplies the numerical solution and at the same time
makes the model more realistic.
The mathematical model is solved by assuming an initial value
for the potential at the crevice mouth and solving the appropriate
electrochemical equations for the crevice and bold surface separately. (The numerical solution of the electrochemical equations is
described below.) If the total cathodic current produced along the
bold surface is not equal in magnitude to the total anodic current
produced in the crevice, the potential at the crevice mouth is modied. In addition, the potential at the outer edge of the cathodic
region must be equal to the corrosion potential. By reducing the
value of potential at the crevice mouth all of the surface overpotentials along the bold surface and in the crevice are reduced via
Eqs. (9) and (12). Such a reduction in potential would decrease the
crevice anodic current and increase the bold surface cathodic current. These changes in anodic and cathodic currents are modeled by
Eqs. (7) and (11), respectively. In other words, a potential gradient
along the bold surface and into the crevice is ultimately calculated,
including the potential at the crevice mouth. This potential gradient
balances all transport and electrochemical production or depletion
of species.
The electrochemical equations relevant to the crevice are Eqs.
(69), and (13). These equations were solved in series and then
iterated until convergence was achieved. The electrochemical equations describing the bold surface are Eqs. (1012). These equations
were explicitly solved since the potential at the crevice mouth and
the total anodic current from the crevice are known [12].
3. Results and discussion
The new model presented in this paper is rst validated without incorporating the hydrogen evolution reaction, and then used
to examine critical crevice scaling laws. A subsequent validation
and a number of numerical simulations were then conducted for
the case where the hydrogen evolution reaction is considered to be
signicant after an electric eld has been established within the

Fig. 1. Model validation (without HER).

crevice. These results were compared with the experimental data


of Alavis and Cottis [7].
3.1. Model without HER
The model was validated against the experimental data of Alavi
and Cottis [7]. Alavi and Cottis studied a crevice of 2.5 cm width,
90 10 m gap, and 8 cm depth. A crevice with walls made from
AISI 304 stainless steel was coupled to a cathode made from AISI
304 stainless steel situated in aerated bulk electrolyte under free
corrosion conditions. The electrolyte was 0.6 M NaCl solution that
was at a temperature of 23 1 C. Fig. 1 shows the pH prole predicted by the model presented in this paper compared with the
experimental data of Alavi and Cottis and predictions made by other
published models that were also compared with the experimental
data of Alavi and Cottis [7]. Fig. 1 shows that the model presented
in this paper more closely predicts the pH prole in the corroding
crevice than any other model prediction displayed in Fig. 1. This
result is very similar to the model of Kennell et al. [12] but there is
a steeper concentration gradient near the mouth and a lower pH at
the centre of the crevice.
The new model described in this paper was used to determine
an applicable critical crevice corrosion scaling law. The model was
used to predict the corrosion currents and electric potential elds
for a variety of AISI 304 stainless steel crevices with the same
electrolyte and physical setup as was used for model validation.
Only the prescribed length and gap of the crevice was altered for
each simulation. If the potential within the crevice at any point
at any time was predicted to be below the critical potential, that
crevice was deemed to become active. If the potential at some point
in the crevice was below the critical potential immediately upon
deoxygenation of the crevice, the crevice was deemed active upon
oxygen depletion. Otherwise, an active crevice was deemed active
after incubation. These results can be seen in Fig. 2, where crevice
dimensions predicted to become active after oxygen depletion are
indicated via solid diamond shapes, and crevice dimensions predicted to become active after a period of incubation are indicated
via solid squares. Also shown in Fig. 2 are the crevices dimensions
in which the metal remained in the passive state, indicated as solid
triangles. Also shown in Fig. 2 are shaded zones that represent a
range of crevice dimensions assumed to be in one of the particular categories outlined above. The corrosion processes for many of
these passive crevices were modeled for extended periods of time,
during which the potential drop into the crevice reached a maximum and then decreased due to an increased conductivity of the
electrolyte in the crevice.

4700

G.F. Kennell, R.W. Evitts / Electrochimica Acta 54 (2009) 46964703

Fig. 4. Predicted electric potentials for the Alavi and Cottis [7] experimental parameters.
Fig. 2. Critical aspect ratio for crevice corrosion.

The critical aspect ratio fell between the active crevice and not
active crevice zones shown in Fig. 2. Two scaling factors from the literature were tted to this data by assuming the midpoint between
the active and not active crevice dimensions was the critical dimension for crevice corrosion. These two scaling factors were L/G [23]
and L2 /G [22]. The rst scaling factor tested, L/G, did not t the data.
The second scaling factor, L2 /G, t the data well. In fact, Fig. 2 shows
that the scaling factor L2 /G delimits the active and non-active zones
shown in the gure very well when of L2 /G = 401 cm.
Interesting observations pertaining to cathodic reactions during several of the predictions from Fig. 2 can be made. Several of
the predictions from Fig. 2 show low amounts of metal deposition
at the crevice tip during some periods of the simulation. The predicted deposition is several orders of magnitude smaller than the
amounts that would be deposited for cathodic currents of the order
of the passive current density, which equates to approximately a
total deposition of 3 ng. Fig. 3 shows the predicted net corrosion current for the crevice of 2.5 cm width, 90 10 m gap, and 8 cm depth,
which is the model validation prediction. The gure shows the
development of the net corrosion current along the crevice at different times during the simulation. As the simulation proceeds, the net
anodic currents towards the mouth of the crevice become greater
as do the concentration gradients towards the crevice mouth. This
changes the electric potential gradient and decreases the potential
at the crevice tip due to Ohmic and electrochemical considerations,
as shown in Fig. 4, which is a plot of potential distance along the
crevice. The potential inside the crevice is calculated with respect
to the bold surface. After 5 h the predicted electric potential at the
crevice tip is equal to the reversible potential of the metal at these
conditions. However, as the electric potential drop approximately
half a centimetre into the crevice mouth continues to drop due to

Fig. 3. Predicted net currents for the Alavi and Cottis [7] experimental parameters
without considering the possibility for the hydrogen evolution reaction.

increasing anodic current and Ohmic considerations, the potential


drop approximately 2 cm into the crevice also continues to drop
due to the diffusion potential and changing concentration gradient
of the electrolyte. This combination of potential drops causes the
potential at the tip of the crevice to fall below the reversible potential estimated for this metal, causing very low levels of net metal
deposition. Fig. 3 shows predicted currents for the case involving
low levels of deposition after 9 h. It should be noted that this prediction only considers the cathodic reaction of metal deposition
and other cathodic reactions that may occur preferentially, such as
the hydrogen evolution reaction, are not considered. Therefore, for
the situation not considering the hydrogen evolution reaction, an
important cause of metal deposition is the diffusion potential in
the crevice electrolyte. The discovery of possible net cathodic reactions at the crevice tip highlights the importance of modeling both
the forward and reverse reactions for types of localized corrosion
similar to crevice corrosion.
3.2. Model incorporating HER
Hydrogen evolution would naturally occur prior to the deposition of metal at the crevice tip. However, hydrogen evolution will
only occur after the crevice has become active and considerable
concentration and potential gradients have been established. The
lack of incorporation of the hydrogen evolution reaction during
the examination of the critical crevice scaling laws in the previous section of this paper is not an important factor due to the lack
of established potential drops into the crevice prior to the initiation of active crevice corrosion. Simulations of crevice corrosion
incorporating the hydrogen evolution reaction were conducted for
the analysis of electrochemical phenomena occurring in the crevice
after a prolonged period of active crevice corrosion. The validation
of the model including the hydrogen evolution reaction, shown in
Fig. 5, ts the experimental data slightly worse than the validation
without the hydrogen evolution reaction shown in Fig. 1. However,
the numerical representation of the hydrogen evolution reaction
in the current model allows for a possible explanation of a peculiar phenomenon displayed in the experimental data of Alavi and
Cottis [7].
During the initial period of the Alavi and Cottis [7] experiment the pH at the crevice tip falls to approximately 4.4, then
rises to about 4.6, and then falls for the rest of the experiment.
During this initial period, the pH at other locations in the crevice
decreases continuously. An explanation for this increasedecrease
may be experimental errors/scatter; however, the numerical simulation including the hydrogen evolution reaction provides an
alternate explanation of why this change in pH might have occurred.
Fig. 6 shows the predicted pH in the crevice at various times. It
can be seen that after 0.4 h the pH at the tip of the crevice is

G.F. Kennell, R.W. Evitts / Electrochimica Acta 54 (2009) 46964703

4701

Fig. 7. Predicted net currents for the Alavi and Cottis [7] experimental parameters
considering the possibility for the hydrogen evolution reaction.

approximately 4. The pH then rises to about 4.1, then the pH


decreases for the rest of the numerical simulation. Although the
comparison of predicted pH with the data of Alavi and Cottis [7] is
only in reasonable agreement, no similar increase in pH was seen at
the crevice tip for simulations conducted without considering the
hydrogen evolution reaction. This increase in pH at the crevice tip
may be explained as follows. Upon initiation of active crevice corrosion high levels of anodic current are produced near the crevice
mouth that produce a large iR drop in the solution. The large iR
drop causes the potential at the crevice tip to be reduced relative
to the bold surface. This decrease in potential increases the rate of
hydrogen evolution along the crevice (especially at the crevice tip).
The evolution of hydrogen at the tip causes an increase in the pH at
the tip. The pH continues to rise until a new quasi-steady state is
achieved and the evolution of hydrogen is balanced with the production (via hydrolysis) and diffusion of hydrogen ions into the tip
region. Then, the pH at the crevice tip continues to decrease along
with the decreasing pH in the remainder of the crevice.
It can therefore be surmised that the rise in pH seen experimentally may be due to the hydrogen evolution reaction occurring
only after an increase in the potential drop inside the crevice (due
to iR drop caused by active corrosion). This further justies the
assumption that the hydrogen evolution reaction is only signicant
after active crevice corrosion initiates causing larger potential drops
within the crevice. Fig. 7 shows the net current at the crevice wall
with depth into the crevice. At the early time periods of 0.5 h and 2 h
corrosion is predicted to occur along the entire length of the crevice;
however, the current emanating from the crevice wall closer to the

mouth increases from 0.5 h to 2 h, causing a larger potential drop,


and reducing the current emanating from the crevice tip (due to
Ohmic considerations). Fig. 7 shows that after 5 h the predicted
anodic current at the crevice wall closer to the mouth has increased,
whilst a net anodic current is no longer predicted to occur towards
the tip. Net cathodic reactions are predicted to occur towards the tip
due mainly to the hydrogen evolution reaction. By the 9th hour the
anodic current towards the crevice mouth is predicted to continue
increasing whilst the cathodic current towards the tip is predicted
to increase in magnitude over a smaller area of the crevice.
Hydrogen evolution had originally been neglected in the model
presented in this paper because no hydrogen evolution was
reported by Alavi and Cottis [7] for their stainless steel crevice
experiment. The maximum current due to the hydrogen evolution
reaction predicted at the crevice tip during the rst 10 simulated hours was approximately 1 108 A/cm2 . This corresponds to
approximately 5 107 mL/min of hydrogen produced in the deepest part of the crevice, where the crevice volume is 1.8 103 mL.
This amount of hydrogen production may go unobserved, but has
an impact on the numerical simulation of the crevice pH. Also, the
evolution of hydrogen at the tip may be the cause of the steady state
concentration gradient of hydrogen ions in the experimental data
of Alavi and Cottis [7]. This pH gradient towards the crevice tip can
be seen in Fig. 5.
Due to repeated numerical and experimental indications that
cathodic reactions may occur towards the crevice tip for 304
stainless steel crevices, a theoretical mechanism was constructed
describing a second cathodic area at the crevice tip, in addition to
the main cathodic area on the bold surface exterior to the crevice.
Important aspects from this mechanism are shown in Fig. 8. Fig. 8
shows the main cathodic reactions occurring on the bold surface

Fig. 6. pH predicted by model considering hydrogen evolution compared with the


experimental data of Alavi and Cottis [7].

Fig. 8. Idealized description of second cathodic area at crevice tip.

Fig. 5. Model validation with hydrogen evolution reaction [24,25].

4702

G.F. Kennell, R.W. Evitts / Electrochimica Acta 54 (2009) 46964703

as the reduction of oxygen molecules into hydroxide molecules.


The passage of ions from the cathodic area and bulk electrolyte
into the crevice are illustrated by chloride and hydroxide along the
dotted arrow to the main anodic area close to the crevice mouth.
Electrons are shown travelling from this anodic area to the main
cathodic area exterior to the crevice, and positive charge exiting the
crevice towards the main cathodic area is shown as positive metal
ions along the dotted arrow. Collectively, the transport of these
various charges completes one electrochemical circuit. Electrons
are also shown travelling from the main anodic area close to the
crevice mouth towards the crevice tip, where a minor cathodic area
may occur. The cathodic reactions occurring towards the crevice
tip are illustrated as the hydrogen evolution reaction. This reaction is more likely to occur as a large potential drop becomes
established. This large potential drop may be largely due to the iR
drop caused by the transport of charged ions between the main
cathodic area and main anodic area, but may also be inuenced
by concentration gradients. Completing the electrochemical circuit between the crevice tip cathodic area and the crevice mouth
anodic area is the transport of positive ions towards the crevice tip.
This is illustrated in Fig. 8 as protons along the dotted arrow. It is
unknown if the hydrogen produced in the crevice tip forms small
bubbles and exits the crevice, or if the hydrogen penetrates into
the metal walls. An important reaction occurring in the crevice, the
hydrolization of water by metal ions, is also shown in Fig. 8. This
mechanism of a second cathodic area would explain the steady state
pH proles demonstrated in the experimental data of Alavi and
Cottis [7].

Acknowledgments
The authors would like to thank the Natural Sciences and Engineering Research Council (NSERC) for nancial support and the
University of Saskatchewan for computing facilities.
Appendix A. Nomenclature

C
D
E0
Ecorr
Ecrit
F
i
i
io
ip
K
KMT
Q
R
S
t
T
TS
u
X
z

concentration (mol/m3 )
diffusion coefcient (m2 /s)
reversible potential (V)
corrosion potential (V)
critical potential (V)
Faradays constant (96,487 C/mol)
current density (A/m2 )
current density vector (A/m2 )
exchange current density (A/m2 )
passive current density (A/m2 )
equilibrium constant
Mass transfer coefcient (m/s)
reaction quotient
universal gas constant (8.3145 J/mol K)
source term (mol/m3 s)
time (s)
temperature (K)
Tafel slope (V/decade)
mobility (m2 mol/J s)
extent of reaction
charge number

4. Conclusions
The numerical model of crevice corrosion presented in this
paper (without considering the hydrogen evolution reaction) predicts the pH prole seen in the experimental data of Alavi and Cottis
[7]. This model demonstrates the importance of modeling both the
forward and reverse reactions of metal dissolution when modeling
localized corrosion. When the hydrogen evolution reaction was not
considered it was found that conditions may exist within a corroding crevice that may cause a net deposition of metal at the crevice
tip. This deposition would be due to an overall potential drop caused
by a combination of an iR drop coupled with a diffusion potential
drop. However, this deposition is unlikely since HER would occur
prior to metal deposition.
Simulations were used to nd a critical crevice corrosion scaling law. The scaling law that t the results was found to be L2 /G.
From data predicted by the current model it was observed that
the critical crevice geometric dimensions were described when
L2 /G 401 cm.
Due to the observed importance of cathodic reactions within a
corroding crevice after establishment of a signicant potential drop
into the crevice (due to the iR drop), further simulations were conducted that incorporated the hydrogen evolution reaction within
the crevice. It was predicted that the hydrogen evolution reaction
occurs in the crevice after concentration and potential gradients
have become signicant due to active crevice corrosion. Due to the
delay in the establishment of larger concentration and potential
gradients, the initiation of hydrogen evolution within the crevice is
delayed. This delay was responsible for an increase in the pH at the
crevice tip during the numerical simulation of the Alavi and Cottis
[7] experiment. An actual delayed increase in the pH at the crevice
tip that was recorded in the experimental data of Alavi and Cottis [7]
may be also be caused by these factors. A new mechanistic model
is also developed that illustrates some of the important factors in
crevice corrosion, including the development of a second cathodic
area towards the crevice tip.

Greek letters

charge density (C/m3 )

electric potential

surface overpotential (V)

conductivity (1 m1 )

stoichiometric coefcient
y
distance between bulk uid and bold surface (m)
z
height of control volume (m)
Subscripts/superscripts
a
anode
c
cathode
H
hydrogen evolution reaction
i
chemical species or reaction
j
chemical species
mouth crevice mouth
last Node last node providing net cathodic current on bold surface
O2
oxygen
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]

H.W. Pickering, Corrosion 42 (1986) 125.


A.M. Al-Zahrani, H.W. Pickering, Electrochim. Acta 50 (2005) 3420.
J.W. Oldeld, W.H. Sutton, Br. Corros. J. 13 (1978) 13.
Y. Xu, H.W. Pickering, Proceedings 92. Proc. Symp. Crit. Factors Localized Corros.,
1991 (1992), 389.
M.K. Watson, J. Postlewaite, Corrosion 46 (1990) 522.
J.C. Walton, G. Cragnolino, S.K. Kalandros, Corros. Sci. 38 (1996) 1.
A. Alavi, R.A. Cottis, Corros. Sci. 27 (1987) 443.
G.R. Engelhardt, D.D. MacDonald, P.J. Millett, Corros. Sci. 41 (1999) 2165.
S.P. White, G.J. Weir, N.J. Laycock, Corros. Sci. 42 (2000) 605.
F. Cui, F. Presuel-Moreno, R. Kelly, Corros. Sci. 47 (2005) 2987.
K.L. Heppner, Ph.D. Thesis, University of Saskatchewan, Saskatoon, SK, 2006.
G.F.S. Kennell, R.W. Evitts, K.L. Heppner, Corros. Sci. 50 (2008) 1716.
M.K. Watson, J. Postlewaite, Proc. Corrosion/90, Las Vegas, NV, NACE International, Houston, TX, April 2327, 1990, p. 156:1.
M.K. Watson, J. Postlethwaite, Corros. Sci. 32 (1991) 1253.
K.L. Heppner, R.W. Evitts, J. Postlethwaite, Can. J. Chem. Eng. 80 (2002)849.
K.L. Heppner, R.W. Evitts, J. Postlethwaite, Can. J. Chem. Eng. 80 (2002) 857.

G.F. Kennell, R.W. Evitts / Electrochimica Acta 54 (2009) 46964703


[17] K.L. Heppner, R.W. Evitts, Int. J. Num. Methods Heat & Fluid Flow. 15 (8) (2005)
842.
[18] K.L. Heppner, R.W. Evitts, Corros. Eng. Sci. Technol. (2006) 110.
[19] L.L. Shreir, R.A. Jarman, G.T. Burstein, Corrosion, 3rd ed., ButterworthHeinemann, Oxford, 1994.
[20] J.S. Newman, Electrochemical Systems, 2nd ed., Prentice-Hall, Englewood Cliffs,
NJ, 1991.

[21]
[22]
[23]
[24]
[25]

4703

L.L. Shrier, Corrosion, vols. I and II, Newness Butterworth, London, 1979.
A. Turnbull, J.G.N. Thomas, J. Electrochem. Soc. 129 (1982) 1412.
Y. Xu, H.W. Pickering, J. Electrochem. Soc. 140 (1993) 658.
R.W. Evitts, Ph.D. Thesis, University of Saskatchewan, Saskatoon, SK, 1997.
S.M. Sharland, P.W. Tasker, Corros. Sci. 28 (1988) 603.

You might also like